Since the final episode of hit sitcom Friends first aired in 2004, many fans have clung to the hope of a reunion. Earlier this year, the show’s co-creator Marta Kauffman quashed that idea emphatically: “There will never be a Friends reunion movie,” she told E! News. Could she be any clearer?
But for those still mourning the gang, there is some sort of hope. A team of researchers at the University of Leeds seeks to immortalize popular characters as digital avatars that you might eventually chat with in the way you talk to Siri, Alexa or a similar virtual assistant.
The first digital character? Joey Tribbiani.
As spotted by Prosthetic Knowledge, the team is building a series of algorithms that can recognize and track individual characters and capture their body language, facial expressions and voice. A machine learning tool analyses the show’s script to understand how each character typically puts words together.
The end result is a “virtually immortalized” version of Joey that can generate new sentences in his own style and voice with a mouth moving to speak these new words.
Some of the algorithmically generated sentences Virtual Joey came up with include “Hey Ross do you want me to talk to some lady” and “I like pizza with cheese”.
It’s not just the dialogue that needs some refinement. As it stands, the digitally synthesized mouth that “talks” like Joey is superimposed over original footage of actor Matt LeBlanc in a comically obvious way; robo-Joey falls squarely into theUncanny Valley.
The team – James Charles, Derek Magee and David Hogg from Leeds University’scomputer science department in the UK – plans to expand the technique to other characters and eventually allow the characters to automatically create new scenes where they have entirely new conversations, thus “effectively rendering the person virtually immortal”.
“We plan to improve the rendering of the avatar and extend our model to include interaction with real people and also between avatars,” said the team in a paperoutlining the technique.
Such a system could be used as a “natural interface between human and computer” to put a “face and personality to existing voice-only assistants” such as Apple’s Siri, Microsoft’s Cortana or Amazon’s Alexa, the researchers said.
Siri, how you doin’?